Applications

Impact of New Undersea Capacity on KENET and East Africa

By: Kevin Chege

Date: October 1, 2010

line break image

Until 2009 on this part of the African continent, connectivity to the Internet was via satellite only. That meant speeds that were slow and connectivity that was expensive. In Tanzania, the price of 1 megabit per second (Mbps) of bandwidth cost anywhere from USD 4,000 to USD 12,000 per Mbps; in Kenya and Uganda, the cost was around USD 2,000 to USD 3,500. (The prices represent bandwidth with a 1-to-1 contention ratio.) Cheaper options were available, but they were probably from a shared pool with no guarantee on the bandwidth. It was possible to find some terrestrial fibre for local loop connectivity, but it was not widespread and was primarily by way of the big telecommunications companies in each of the countries.

In 2008, word of undersea connectivity began to spread, but many in the region grew skeptical when the landing dates for the cables kept changing. Finally, three cables arrived in the region: SEACOM and TEAMS, each with 1.28 terabit-per-second capacity, and the EASSy cable, with a capacity of 3.8 terabits. SEACOM, which would land on Kenya’s coast at Mombasa and on Tanzania’s coast at Dar es Salaam, would connect the two countries to London via Marseille; TEAMS would connect Mombasa only to Fujairah in the United Arab Emirates. EASSy, which was to land in nine countries from South Africa to Sudan along Africa’s east coast, was meant primarily to interconnect African countries internally.

In 2009, two cables were completed: SEACOM in June and TEAMS in July. In addition to greater Internet capacity, many in the region were looking forward to lower connectivity prices. SEACOM had a head start by having arranged for its connectivity to be carried all the way to Uganda and Rwanda via Kenya and also by arriving a month earlier than the competition’s. Part of the reason the price of connectivity on the SEACOM cable was more favourable than on the TEAMS cable was that Internet Protocol transit costs in London were much less than they are in Fujairah. In London, the prices that the Kenya Education Network Trust (KENET) received ranged from USD 2 to USD 15 depending on the amount of bulk bandwidth purchased. In Fujairah, transit costs ranged from USD 70 to USD 90 per 1 Mbps, also varying with the amount purchased. Eventually, the TEAMS cable was put into service, though in Kenya there was some fallout among some of the members who owned the TEAMS cable, which resulted in some shares’ being bought out. EASSy was not completed until mid-2010.

Of the three regions, Kenya appears to be the best served in terms of capacity, with all three cables landing at Kenya’s port of Mombasa. In late 2009, the new pricing started reaching consumers. In Kenya, prices dropped to around USD 600 per Mbps. One of the large Internet service providers (ISPs) offered a buy-one-get-four offer, so the price ultimately climbed to USD 150 per Mbps, but only if a consumer purchased 4 Mbps. With a store of unused capacity available, the ISPs could afford to be generous. Some ISPs had activated 10 gigabits per second (Gbps) of capacity, but only 2.5 Gbps were typically in use.

At, KENET, we expected universities to be using as much as 5 or 10 times their previous usage, but despite the faster speeds to the Internet (up from the previous 700 to 800 milliseconds (ms) round-trip time (RTT) to 180 to 300 ms RTT), they were not. We discovered that the low usage was due primarily to poor internal networks that were sufficient for VSAT (very-small-aperture terminal) but not for fibre. In addition, the networks were not well set up, so high speeds were observed only at the edge of the network—near the router gateway to KENET—and there was a low ratio of personal computers to users. (See http://eready.kenet.or.ke.)

We tried to solve the problem of poor internal infrastructure by offering more consultations and trainings. In March 2010, a workshop was conducted in partnership with the Network Startup Resource Center (NSRC) and the University of Oregon covering campus network design, which is the foundation of developing a robust, high-performance National Research and Education Network (NREN). The main objectives of the workshop were to train university network engineering staff on how to develop and implement a strategic design plan for their campus networks, with particular emphasis on layers 1 and 2, and to strengthen the KENET technical community (human network) in developing KENET’s cyberinfrastructure.

Funding and support for the training came from the National Science Foundation (NSF) and the Partnership for Higher Education in Africa (PHEA) via the NSRC. As an additional benefit of the training, the NSRC donated 2.5 tons of technical reference books and networking equipment (routers, wireless equipment and gigabit network switches) that have been distributed to the universities for live deployment following the training. This equipment is making a significant impact in improving the functionality of the respective campus networks, which helps to achieve KENET’s goal of getting more faculty members and students using the Internet for research and education activities.

In addition, more emphasis was put on bandwidth management and optimization because even though there was more capacity, it was not being shared properly: one or two users were still able to consume the majority of the available capacity available to the universities. KENET is undertaking an effort to extend this programme further and to provide more online information on the topic, including how-tos and online training materials hosted at KENET (see http://bmo.kenet.or.ke and http://training.kenet.or.ke). The information is also available to the neighbouring national research and education networks that form part of KENET’s mailing lists on the topic.

KENET had previously signed a two-year VSAT contract with a supplier for 200 Mbps of capacity. Though the contract became effective in January 2009, the resources that were to be made available by means of a World Bank grant to improve Internet capacity at the universities got delayed. Regardless, KENET managed to negotiate with the provider for half of the anticipated capacity with fibre—meaning, KENET would remain active, with 100 Mbps VSAT and 100 Mbps equivalent by way of fibre when it became active. Thus, KENET would have 350-Mbps capacity on fibre and 100-Mbps capacity on VSAT plus an additional 155 Mbps via a KENET-owned link to the UbuntuNet router in London connecting KENET to GÉANT and allowing for peering via UbuntuNet.

Since that expansion, KENET went from 12 Mbps in 2008 on VSAT to 450 Mbps on VSAT plus fibre in 2009, to more than 600 Mbps combined capacity in 2010—all in a span of 18 months. Usage has peaked, and the need for additional capacity is anticipated due to users’ increased awareness of better speeds, which leads to increased use of the Internet for reading newspapers online, conducting research, streaming video, and visiting social networking sites.

In response to rising demand, KENET has increased its infrastructure for core services like DNS and email, which have become more essential than ever. There is also more demand for online services, such as email filtering and Web hosting as more of our universities increase their online communication facilities. KENET is building both a network operations centre and a data centre to facilitate deployment of additional services, such as Web-based conferencing, caching, backup services, and storage.

Today in Kenya, with more users online than ever, more companies than ever are using e-marketing. However, there are issues related to online security—particularly with recent identification of Kenya as the country most hit by viruses, a problem that also affects the country’s universities. KENET is planning to address that problem by examining its policies and increasing its security training in the coming year. KENET is also seeing more peer-to-peer traffic, a concern that is also being addressed through education and training. At this time, not all of the universities are in a position to control such problems, so the idea is to empower them to be able to control and manage their capacities. For even our most remote universities, VSAT is no longer an option, and since fibre is not available throughout the country, we are hard-pressed to find viable solutions that include last-mile-high-capacity radios to backhaul to the nearest point of fibre.

In closing, the impact of increased capacity is more demand for online, high-quality services as well as a big drop in the cost of capacity. Universities have been able to scale up to as much as 10 times their previous capacity—from 10 Mbps on VSAT to 60 Mbps on fibre. The biggest constraint for Universities in Kenya and East Africa is the last-mile reach and cost, which is slowly improving. However, the universities are working on upgrading their internal networks and KENET is working on deploying wireless networks in the campuses. The current cost is now around USD 280 per Mbps with the price set to fall even further in 2011 as more ISPs and KENET drop their VSAT contracts.

It is big news when an undersea cable is cut, as was evident when the main cable, SEACOM, experienced a repeater failure for two weeks in July 2010. KENET was forced to purchase restoration capacity in order to support our VSAT redundancy capacity, which was too slow for users. Our capacity of 606 Mbps is now fully used up. In order to meet ever-growing demand, we are planning to activate more capacity via the TEAMS cable before the end of 2010 and we anticipate our capacity in January 2011 to be 750 Mbps and 1.2 Gbps later that year.

Kevin Chege is network manager of Kenya Education Network Trust

This article was posted on 31 January 2011

No Comments to Show

Leave a Reply

Your email address will not be published. Required fields are marked *